A Direct Comparison of Remote Sensing Approaches for High-Throughput Phenotyping in Plant Breeding

Translating High-Throughput Phenotyping into Genetic Gain

Inability to efficiently implement high-throughput field phenotyping is increasingly perceived as a key component that limits genetic gain in breeding programs. Field phenotyping must be integrated into a wider context than just choosing the correct selection traits, deployment tools, evaluation platforms, or basic data-management methods. Phenotyping means more than conducting such activities in a resource-efficient manner; it also requires appropriate trial management and spatial variability handling, definition of key constraining conditions prevalent in the target population of environments, and the development of more comprehensive data management, including crop modeling. This review will provide a wide perspective on how field phenotyping is best implemented. It will also outline how to bridge the gap between breeders and 'phenotypers' in an effective manner.

Genome-Wide Association Analyses Identify QTL Hotspots for Yield and Component Traits in Durum Wheat Grown under Yield Potential, Drought, and Heat Stress Environments

Understanding the genetic bases of economically important traits is fundamentally important in enhancing genetic gains in durum wheat. In this study, a durum panel of 208 lines (comprised of elite materials and exotics from the International Maize and Wheat Improvement Center gene bank) were subjected to genome wide association study (GWAS) using 6,211 DArTseq single nucleotide polymorphisms (SNPs). The panel was phenotyped under yield potential (YP), drought stress (DT), and heat stress (HT) conditions for 2 years. Mean yield of the panel was reduced by 72% (to 1.64 t/ha) under HT and by 60% (to 2.33 t/ha) under DT, compared to YP (5.79 t/ha). Whereas, the mean yield of the panel under HT was 30% less than under DT. GWAS identified the largest number of significant marker-trait associations on chromosomes 2A and 2B with p-values 10-06 to 10-03 and the markers from the whole study explained 7-25% variation in the traits. Common markers were identified for stress tolerance indices: stress susceptibility index, stress tolerance, and stress tolerance index estimated for the traits under DT (82 cM on 2B) and HT (68 and 83 cM on 3B; 25 cM on 7A). GWAS of irrigated (YP and HT combined), stressed (DT and HT combined), combined analysis of three environments (YP + DT + HT), and its comparison with trait per se and stress indices identified QTL hotspots on chromosomes 2A (54-70 cM) and 2B (75-82 cM). This study enhances our knowledge about the molecular markers associated with grain yield and its components under different stress conditions. It identifies several marker-trait associations for further exploration and validation for marker-assisted breeding.

Hyperspectral reflectance as a tool to measure biochemical and physiological traits in wheat

Improving photosynthesis to raise wheat yield potential has emerged as a major target for wheat physiologists. Photosynthesis-related traits, such as nitrogen per unit leaf area (Narea) and leaf dry mass per area (LMA), require laborious, destructive, laboratory-based methods, while physiological traits underpinning photosynthetic capacity, such as maximum Rubisco activity normalized to 25 °C (Vcmax25) and electron transport rate (J), require time-consuming gas exchange measurements. The aim of this study was to assess whether hyperspectral reflectance (350-2500 nm) can be used to rapidly estimate these traits on intact wheat leaves. Predictive models were constructed using gas exchange and hyperspectral reflectance data from 76 genotypes grown in glasshouses with different nitrogen levels and/or in the field under yield potential conditions. Models were developed using half of the observed data with the remainder used for validation, yielding correlation coefficients (R2 values) of 0.62 for Vcmax25, 0.7 for J, 0.81 for SPAD, 0.89 for LMA, and 0.93 for Narea, with bias <0.7%. The models were tested on elite lines and landraces that had not been used to create the models. The bias varied between -2.3% and -5.5% while relative error of prediction was similar for SPAD but slightly greater for LMA and Narea.

Correction to: Strategic crossing of biomass and harvest index-source and sink-achieves genetic gains in wheat

[This corrects the article DOI: 10.1007/s10681-017-2040-z.].

Plant Science View on Biohybrid Development

Biohybrid consists of a living organism or cell and at least one engineered component. Designing robot-plant biohybrids is a great challenge: it requires interdisciplinary reconsideration of capabilities intimate specific to the biology of plants. Envisioned advances should improve agricultural/horticultural/social practice and could open new directions in utilization of plants by humans. Proper biohybrid cooperation depends upon effective communication. During evolution, plants developed many ways to communicate with each other, with animals, and with microorganisms. The most notable examples are: the use of phytohormones, rapid long-distance signaling, gravity, and light perception. These processes can now be intentionally re-shaped to establish plant-robot communication. In this article, we focus on plants physiological and molecular processes that could be used in bio-hybrids. We show phototropism and biomechanics as promising ways of effective communication, resulting in an alteration in plant architecture, and discuss the specifics of plants anatomy, physiology and development with regards to the bio-hybrids. Moreover, we discuss ways how robots could influence plants growth and development and present aims, ideas, and realized projects of plant-robot biohybrids.

Unmanned Aerial Vehicle Remote Sensing for Field-Based Crop Phenotyping: Current Status and Perspectives

Phenotyping plays an important role in crop science research; the accurate and rapid acquisition of phenotypic information of plants or cells in different environments is helpful for exploring the inheritance and expression patterns of the genome to determine the association of genomic and phenotypic information to increase the crop yield. Traditional methods for acquiring crop traits, such as plant height, leaf color, leaf area index (LAI), chlorophyll content, biomass and yield, rely on manual sampling, which is time-consuming and laborious. Unmanned aerial vehicle remote sensing platforms (UAV-RSPs) equipped with different sensors have recently become an important approach for fast and non-destructive high throughput phenotyping and have the advantage of flexible and convenient operation, on-demand access to data and high spatial resolution. UAV-RSPs are a powerful tool for studying phenomics and genomics. As the methods and applications for field phenotyping using UAVs to users who willing to derive phenotypic parameters from large fields and tests with the minimum effort on field work and getting highly reliable results are necessary, the current status and perspectives on the topic of UAV-RSPs for field-based phenotyping were reviewed based on the literature survey of crop phenotyping using UAV-RSPs in the Web of Science™ Core Collection database and cases study by NERCITA. The reference for the selection of UAV platforms and remote sensing sensors, the commonly adopted methods and typical applications for analyzing phenotypic traits by UAV-RSPs, and the challenge for crop phenotyping by UAV-RSPs were considered. The review can provide theoretical and technical support to promote the applications of UAV-RSPs for crop phenotyping.

Intercomparison of Unmanned Aerial Vehicle and Ground-Based Narrow Band Spectrometers Applied to Crop Trait Monitoring in Organic Potato Production

Vegetation properties can be estimated using optical sensors, acquiring data on board of different platforms. For instance, ground-based and Unmanned Aerial Vehicle (UAV)-borne spectrometers can measure reflectance in narrow spectral bands, while different modelling approaches, like regressions fitted to vegetation indices, can relate spectra with crop traits. Although monitoring frameworks using multiple sensors can be more flexible, they may result in higher inaccuracy due to differences related to the sensors characteristics, which can affect information sampling. Also organic production systems can benefit from continuous monitoring focusing on crop management and stress detection, but few studies have evaluated applications with this objective. In this study, ground-based and UAV spectrometers were compared in the context of organic potato cultivation. Relatively accurate estimates were obtained for leaf chlorophyll (RMSE = 6.07 µg·cm⁻²), leaf area index (RMSE = 0.67 m²·m⁻²), canopy chlorophyll (RMSE = 0.24 g·m⁻²) and ground cover (RMSE = 5.5%) using five UAV-based data acquisitions, from 43 to 99 days after planting. These retrievals are slightly better than those derived from ground-based measurements (RMSE = 7.25 µg·cm⁻², 0.85 m²·m⁻², 0.28 g·m⁻² and 6.8%, respectively), for the same period. Excluding observations corresponding to the first acquisition increased retrieval accuracy and made outputs more comparable between sensors, due to relatively low vegetation cover on this date. Intercomparison of vegetation indices indicated that indices based on the contrast between spectral bands in the visible and near-infrared, like OSAVI, MCARI2 and CI_g provided, at certain extent, robust outputs that could be transferred between sensors. Information sampling at plot level by both sensing solutions resulted in comparable discriminative potential concerning advanced stages of late blight incidence. These results indicate that optical sensors, and their integration, have great potential for monitoring this specific organic cropping system.

Field-based high throughput phenotyping rapidly identifies genomic regions controlling yield components in rice

To ensure food security in the face of population growth, decreasing water and land for agriculture, and increasing climate variability, crop yields must increase faster than the current rates. Increased yields will require implementing novel approaches in genetic discovery and breeding. Here we demonstrate the potential of field-based high throughput phenotyping (HTP) on a large recombinant population of rice to identify genetic variation underlying important traits. We find that detecting quantitative trait loci (QTL) with HTP phenotyping is as accurate and effective as traditional labor-intensive measures of flowering time, height, biomass, grain yield, and harvest index. Genetic mapping in this population, derived from a cross of an modern cultivar (IR64) with a landrace (Aswina), identified four alleles with negative effect on grain yield that are fixed in IR64, demonstrating the potential for HTP of large populations as a strategy for the second green revolution.

Assessment of Vegetation Indices Derived by UAV Imagery for Durum Wheat Phenotyping under a Water Limited and Heat Stressed Mediterranean Environment

There is growing interest for using Spectral Vegetation Indices (SVI) derived by Unmanned Aerial Vehicle (UAV) imagery as a fast and cost-efficient tool for plant phenotyping. The development of such tools is of paramount importance to continue progress through plant breeding, especially in the Mediterranean basin, where climate change is expected to further increase yield uncertainty. In the present study, Normalized Difference Vegetation Index (NDVI), Simple Ratio (SR) and Green Normalized Difference Vegetation Index (GNDVI) derived from UAV imagery were calculated for two consecutive years in a set of twenty durum wheat varieties grown under a water limited and heat stressed environment. Statistically significant differences between genotypes were observed for SVIs. GNDVI explained more variability than NDVI and SR, when recorded at booting. GNDVI was significantly correlated with grain yield when recorded at booting and anthesis during the 1st and 2nd year, respectively, while NDVI was correlated to grain yield when recorded at booting, but only for the 1st year. These results suggest that GNDVI has a better discriminating efficiency and can be a better predictor of yield when recorded at early reproductive stages. The predictive ability of SVIs was affected by plant phenology. Correlations of grain yield with SVIs were stronger as the correlations of SVIs with heading were weaker or not significant. NDVIs recorded at the experimental site were significantly correlated with grain yield of the same set of genotypes grown in other environments. Both positive and negative correlations were observed indicating that the environmental conditions during grain filling can affect the sign of the correlations. These findings highlight the potential use of SVIs derived by UAV imagery for durum wheat phenotyping under low yielding Mediterranean conditions.

Comparison of ground cover estimates from experiment plots in cotton, sorghum and sugarcane based on images and ortho-mosaics captured by UAV

Ground cover is an important physiological trait affecting crop radiation capture, water-use efficiency and grain yield. It is challenging to efficiently measure ground cover with reasonable precision for large numbers of plots, especially in tall crop species. Here we combined two image-based methods to estimate plot-level ground cover for three species, from either an ortho-mosaic or undistorted (i.e. corrected for lens and camera effects) images captured by cameras using a low-altitude unmanned aerial vehicle (UAV). Reconstructed point clouds and ortho-mosaics for the whole field were created and a customised image processing workflow was developed to (1) segment the 'whole-field' datasets into individual plots, and (2) 'reverse-calculate' each plot from each undistorted image. Ground cover for individual plots was calculated by an efficient vegetation segmentation algorithm. For 79% of plots, estimated ground cover was greater from the ortho-mosaic than from images, particularly when plants were small, or when older/taller in large plots. While there was a good agreement between the ground cover estimates from ortho-mosaic and images when the target plot was positioned at a near-nadir view near the centre of image (cotton: R2=0.97, sorghum: R2=0.98, sugarcane: R2=0.84), ortho-mosaic estimates were 5% greater than estimates from these near-nadir images. Because each plot appeared in multiple images, there were multiple estimates of the ground cover, some of which should be excluded, e.g. when the plot is near edge within an image. Considering only the images with near-nadir view, the reverse calculation provides a more precise estimate of ground cover compared with the ortho-mosaic. The methodology is suitable for high throughput phenotyping for applications in agronomy, physiology and breeding for different crop species and can be extended to provide pixel-level data from other types of cameras including thermal and multi-spectral models.